Theodosios Goudas

Argument Extraction from News, Blogs, and Social Media

Argument extraction is the task of identifying arguments, along with their components in text. Arguments can be usually decomposed into a claim and one or more premises justifying it. Among the novel aspects of this work is the thematic domain itself which relates to Social Media, in contrast to traditional research in the area, which concentrates mainly on law documents and scientific publications. The huge increase of social media communities, along with their user tendency to debate, makes the identification of arguments in these texts a necessity. Argument extraction from Social Media is more challenging because texts may not always contain arguments, as is the case of legal documents or scientific publications usually studied. In addition, being less formal in nature, texts in Social Media may not even have proper syntax or spelling. This paper presents a two-step approach for argument extraction from social media texts. During the first step, the proposed approach tries to classify the sentences into “sentences that contain arguments” and “sentences that don’t contain arguments”. In the second step, it tries to identify the exact fragments that contain the premises from the sentences that contain arguments, by utilizing conditional random fields. The results exceed significantly the base line approach, and according to literature, are quite promising.

Ratsnake: A Versatile Image Annotation Tool with Application to Computer-Aided Diagnosis

Image segmentation and annotation are key components of image-based medical computer-aided diagnosis (CAD) systems. In this paper we present Ratsnake, a publicly available generic image annotation tool providing annotation efficiency, semantic awareness, versatility, and extensibility, features that can be exploited to transform it into an effective CAD system. In order to demonstrate this unique capability, we present its novel application for the evaluation and quantification of salient objects and structures of interest in kidney biopsy images. Accurate annotation identifying and quantifying such structures in microscopy images can provide an estimation of pathogenesis in obstructive nephropathy, which is a rather common disease with severe implication in children and infants. However a tool for detecting and quantifying the disease is not yet available. A machine learning-based approach, which utilizes prior domain knowledge and textural image features, is considered for the generation of an image force field customizing the presented tool for automatic evaluation of kidney biopsy images. The experimental evaluation of the proposed application of Ratsnake demonstrates its efficiency and effectiveness and promises its wide applicability across a variety of medical imaging domains.

A novel robust approach for handling illumination changes in video segmentation

In this work we propose a novel algorithm for foreground segmentation in video sequences that achieves better accuracy, while maintaining low complexity and allows real time execution. The proposed algorithm combines selected features from a number of well-established and state of the art algorithms, such as the Gaussian mixture models, the Self Organizing Maps and the illumination sensitive method. The presented methodology is capable of efficiently handling sudden light changes, both from natural and multiple artificial light sources, which may have caused erroneous segmentation for other algorithms. Comparative results are presented utilizing user-defined ground truth segmentation for two different types of indoor video sequences, one of which was obtained by a hemispheric omnidirectional dioptric (fish-eye) camera, with and without illumination changes and the second by a plain camera. The behavior of the algorithm with respect to its controlling parameters is investigated and its computational burden is studied. A segmentation algorithm is proposed for video sequences with illumination changes.The proposed algorithm maintains low complexity and allows real time executionThe proposed algorithm combines features from state of the art algorithmsComparative results are presented for videos obtained by omnidirectional and by plain camera

Fisheye camera modeling for human segmentation refinement in indoor videos

In this paper, we concentrate on refining the results of segmenting human presence from indoors videos acquired by a fisheye camera, using a 3D mathematical model of the camera. The model has been calibrated according to the specific indoor environment that is being monitored. Human segmentation is implemented using a standard established technique. The fisheye camera used for video acquisition is modeled using a spherical element, while the parameters of the camera model are determined only once, using the correspondence of a number of user-defined landmarks, both in real world coordinates and on the acquired video frame. Subsequently, each pixel of the video frame is inversely mapped to the direction of view in the real world and the relevant data are stored in look-up tables for very fast utilization in real-time video processing. The proposed fisheye camera model enables the inference of possible real world positions of a segmented cluster of pixels in the video frame. In this work, we utilize the constructed camera model to achieve a simple geometric reasoning that corrects gaps and mistakes of the human figure segmentation. Initial results are also presented for a small number of video sequences, which prove the efficiency of the proposed method.

Argument Extraction from News, Blogs, and the Social Web

Argument extraction is the task of identifying arguments, along with their components in text. Arguments can be usually decomposed into a claim and one or more premises justifying it. Among the novel aspects of this work is the thematic domain itself which relates to Social Media, in contrast to traditional research in the area, which concentrates mainly on law documents and scientific publications. The huge increase of social media communities, along with their user tendency to debate, makes the identification of arguments in these texts a necessity. Argument extraction from Social Media is more challenging because texts may not always contain arguments, as is the case of legal documents or scientific publications usually studied. In addition, being less formal in nature, texts in Social Media may not even have proper syntax or spelling. This paper presents a two-step approach for argument extraction from social media texts. During the first step, the proposed approach tries to classify the sentences into “sentences that contain arguments” and “sentences that don’t contain arguments”. In the second step, it tries to identify the exact fragments that contain the premises from the sentences that contain arguments, by utilizing conditional random fields. The results exceed significantly the base line approach, and according to literature, are quite promising.

A Collaborative Biomedical Image-Mining Framework: Application on the Image Analysis of Microscopic Kidney Biopsies

The analysis and characterization of biomedical image data is a complex procedure involving several processing phases, such as data acquisition, preprocessing, segmentation, feature extraction, and classification. The proper combination and parameterization of the utilized methods are heavily relying on the given image dataset and experiment type. They may thus necessitate advanced image processing and classification knowledge and skills from the side of the biomedical expert. In this study, an application, exploiting web services and applying ontological modeling, is presented, to enable the intelligent creation of image-mining workflows. The described tool can be directly integrated to the RapidMiner, Taverna or similar workflow management platforms. A case study dealing with the creation of a sample workflow for the analysis of kidney biopsy microscopy images is presented to demonstrate the functionality of the proposed framework.

Cancer cells detection and pathology quantification utilizing image analysis techniques

This paper presents an advanced image analysis tool for the accurate and fast characterization and quantification of cancer and apoptotic cells in microscopy images utilizing adaptive thresholding and a Support Vector Machines classifier. The segmentation results are also enhanced through a Majority Voting and a Watershed technique. The proposed tool was evaluated by experts on breast cancer images and the reported results were accurate and reproducible.

A novel hybrid approach for human silhouette segmentation

In this work we propose a novel algorithm for human silhouette segmentation, which combines characteristics from a number of well established and state of the art algorithms, such as the Gaussian mixture models, the Self Organizing Maps and the Illumination Sensitive method. The proposed algorithm is evaluated against user-defined ground truth segmentation for two different types of indoor video sequences, one of which was obtained by a hemispheric camera. The behavior of the algorithm with respect to its controlling parameters is investigated and its computational burden is studied.

An Advanced Image Analysis Tool for the Quantification and Characterization of Breast Cancer in Microscopy Images

The paper presents an advanced image analysis tool for the accurate and fast characterization and quantification of cancer and apoptotic cells in microscopy images. The proposed tool utilizes adaptive thresholding and a Support Vector Machines classifier. The segmentation results are enhanced through a Majority Voting and a Watershed technique, while an object labeling algorithm has been developed for the fast and accurate validation of the recognized cells. Expert pathologists evaluated the tool and the reported results are satisfying and reproducible.

Advanced cancer cell characterization and quantification of microscopy images

In this paper we present an advanced image analysis tool for the accurate characterization and quantification of cancer and apoptotic cells in microscopy images. Adaptive thresholding and Support Vector Machines classifiers were utilized for this purpose. The segmentation results are improved through the application of morphological operators such as Majority Voting and a Watershed technique. The proposed tool was evaluated on breast cancer images by medical experts and the results were accurate and reproducible.